Proper Loans Explanatory Data Analysis
Wafic Fahme
Introduction
Why Prosper:
Prosper is a leading P2P in the lending industry and it attracted me most
because I work in the financial sector and companies like Lending Club and
Prosper are interesting to us as we read a lot about them in the media.
Univariate Plots
## [1] 113937 81In the prosper data set provided we have 81 variables and 113,937
observation.
We can see that the APR every borrower of the 114,000 loans we have payed.
It looks very similar to APR in shape and distribution
Almost half of the loans are still in process, and around 4,000 of the loans have been completed, but we can see almost 11,000 charged off their loans and
almost 5,000 borrowers defaulted on their loans.
##
## Cancelled Chargedoff Completed
## 5 11992 38074
## Current Defaulted FinalPaymentInProgress
## 56576 5018 205
## Past Due (>120 days) Past Due (1-15 days) Past Due (16-30 days)
## 16 806 265
## Past Due (31-60 days) Past Due (61-90 days) Past Due (91-120 days)
## 363 313 304Number of loans in the every status, we have around 12,000 loans of the
114,000 observations charged off
We can seee that the majority of loans (around 80,000) fall in the middle
which gets us to assume that usually loan amounts has to be small to be paid in such a short term
Loan Terms
##
## 12 36 60
## 1614 87778 24545We have 3 groups of borrowers here, with around 88,000 in the 36 month period
(3yrs) and period and the second popular is 60 month period (5yrs) with 25,000
borrowers and least is the 12 months which I assume is for borrowers of small
amounts
We can see spikes at some specific loan amount like 10,000, 20,000 and 25,000
Lets break it see the amount more clear
Loan Original Amount
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 1000 4000 6500 8337 12000 35000We can see that the median of a loan at prosper is 6,500 which confirms our
earlier expectations of why the loan terms are mostly 36 month period.
The number of borrowers in the Loan Original Amount is right skewed, although we can see spikes at amounts:
- $4,000 (15,000 borrowers) - $5,000 (7,500 borrowers) - $10,000 (13,000 borrowers) - $15,000 (15,000 borrowers) - $20,000 (3,000 borrowers) - $30,000 (3,000 borrowers) This makes sense because people will take a loan for 15,000 if they want around
14,000 and when the figures become bigger we see less and less variations in the amounts borrowed unlike the smaller amounts when 1,000 makes a difference
for someome wanting to borrow 2,000.
If we look at years 2008-2009, we will notice that in 2008 the growth of the
number of borrowers stall and then dramatically dropped due to the housing
crisis, but then it started growing in a fast pace. In 4 years the number of
loans went from 5,000 in 2010 to 35,000 loans in 2013.
Around 60,000 borrowers salary ranges between $25,000 and $75,000 but the
surprise is in the numbers of borrowers whom salary is $100,000 and above
whom are around 17,000 borrowers, I wonder how much they borrowed especially
that the amount borrowed median is $6,500, are they the borrowers of the
$30,000 we saw earlier.
We can see that the highest state in terms of prosper borowers is CA which
is expected because such P2P startups are still in the early adopters stage
and usually states like California tends to have have cultures that are more
adoptive of online and new tech ideas.
Almost above 90,000 of the borrowers are either employed and very little
are un-employed or retired, I wonder how much they have borrowed for how long
We can see that the majority of borrowers have a monthly loan payment
less than $300, even 30,000 of the borrowers pay as low as arounf $150 per
month
After factoring and leveling the data, we have almost 90,000 borrowes of our
113,000 has no deliquencies whatsoever with around 12,000 whom have 1 account
delinquent.
Univariate Analysis
What is the structure of your dataset?
The data set has 81 variables, but I have chosen these to work with for my EDA.
ListingCreationDate: The date the listing was created.
Term: The length of the loan expressed in months.
LoanStatus: The current status of the loan: Cancelled, Chargedoff, Completed, Current, Defaulted, FinalPaymentInProgress, PastDue. The PastDue status will be accompanied by a delinquency bucket.
BorrowerAPR: The Borrower’s Annual Percentage Rate (APR) for the loan.
BorrowerRate: The Borrower’s interest rate for this loan.
ProsperRating (Alpha): The Prosper Rating assigned at the time the listing was created between AA - HR. Applicable for loans originated after July 2009.
ProsperScore: A custom risk score built using historical Prosper data. The score ranges from 1-10, with 10 being the best, or lowest risk score.
Applicable for loans originated after July 2009.CreditScoreRangeUpper: The upper value representing the range of the borrower’s credit score as provided by a consumer credit rating agency.
AmountDelinquent: Dollars delinquent at the time the credit profile was pulled.
BankcardUtilization: The percentage of available revolving credit that is utilized at the time the credit profile was pulled.
DebtToIncomeRatio: The debt to income ratio of the borrower at the time the credit profile was pulled. This value is Null if the debt to income ratio is not available. This value is capped at 10.01 (any debt to income ratio larger than 1000% will be returned as 1001%).
IncomeRange: The income range of the borrower at the time the listing was created.
StatedMonthlyIncome: The monthly income the borrower stated at the time the listing was created.
LoanOriginalAmount: The origination amount of the loan.
Investors: The number of investors that funded the loan.
MonthlyLoanPayment The scheduled monthly loan payment.
BorrowerState: The two letter abbreviation of the state of the address of the borrower at the time the Listing was created.
Create a new dataframe with the above variables
## [1] "Term" "LoanStatus"
## [3] "BorrowerAPR" "BorrowerRate"
## [5] "ListingKey" "MonthlyLoanPayment"
## [7] "ProsperRating..Alpha." "ProsperScore"
## [9] "BorrowerState" "CreditScoreRangeUpper"
## [11] "AmountDelinquent" "BankcardUtilization"
## [13] "DebtToIncomeRatio" "IncomeRange"
## [15] "StatedMonthlyIncome" "LoanOriginalAmount"
## [17] "Investors" "BorrowersGroups"
## [19] "ListingCreationYear"## [1] "Current" "Completed"
## [3] "Chargedoff" "Cancelled"
## [5] "Defaulted" "FinalPaymentInProgress"
## [7] "Past Due (1-15 days)" "Past Due (16-30 days)"
## [9] "Past Due (31-60 days)" "Past Due (61-90 days)"
## [11] "Past Due (91-120 days)" "Past Due (>120 days)"I will group Current, Completed, Past Due (1-15 days), Past Due (16-30 days)
and “FinalPaymentInProgress”
as Non-Delinquent because a borrower who is only few days late on his or her
payment might have reasons that are quite random an not covered in the data
we have. Then I will group all the 8 levels into delinquent borrowers
because for example a borrower whom is 90 days past due is probably gonna cancel or default on his loan and I believe they might also be delinquent
elsewhere as we will try to investigate via correlating the status with the number of delinquencies a borrower has.
## 'data.frame': 113937 obs. of 19 variables:
## $ Term : Factor w/ 3 levels "12","36","60": 2 2 2 2 2 3 2 2 2 2 ...
## $ LoanStatus : Factor w/ 12 levels "Current","Completed",..: 2 1 2 1 1 1 1 1 1 1 ...
## $ BorrowerAPR : num 0.165 0.12 0.283 0.125 0.246 ...
## $ BorrowerRate : num 0.158 0.092 0.275 0.0974 0.2085 ...
## $ ListingKey : chr "1021339766868145413AB3B" "10273602499503308B223C1" "0EE9337825851032864889A" "0EF5356002482715299901A" ...
## $ MonthlyLoanPayment : num 330 319 123 321 564 ...
## $ ProsperRating..Alpha.: Factor w/ 7 levels "AA","A","B","C",..: NA 2 NA 2 5 3 6 4 1 1 ...
## $ ProsperScore : num NA 7 NA 9 4 10 2 4 9 11 ...
## $ BorrowerState : chr "CO" "CO" "GA" "GA" ...
## $ CreditScoreRangeUpper: int 659 699 499 819 699 759 699 719 839 839 ...
## $ AmountDelinquent : num 472 0 NA 10056 0 ...
## $ BankcardUtilization : num 0 0.21 NA 0.04 0.81 0.39 0.72 0.13 0.11 0.11 ...
## $ DebtToIncomeRatio : num 0.17 0.18 0.06 0.15 0.26 0.36 0.27 0.24 0.25 0.25 ...
## $ IncomeRange : Factor w/ 7 levels "Not employed",..: 4 5 NA 4 7 7 4 4 4 4 ...
## $ StatedMonthlyIncome : num 3083 6125 2083 2875 9583 ...
## $ LoanOriginalAmount : int 9425 10000 3001 10000 15000 15000 3000 10000 10000 10000 ...
## $ Investors : int 258 1 41 158 20 1 1 1 1 1 ...
## $ BorrowersGroups : Factor w/ 2 levels "Non-Delinquent",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ ListingCreationYear : chr "2007" "2014" "2007" "2012" ...What is/are the main feature(s) of interest in your dataset?
I was suprised to see a varying APR levels for every borrower and I was
curious to undertsand why the difference.
I want investigate why APR levels changes from loan to another, what are the
factors that get me as a borrower to have a lower APR at prosper. I hope the
variables will help me understand the reason beind the difference in rates.
### What other features in the dataset do you think will help support your
investigation into your feature(s) of interest? We will look into the financial status of the borrower thus the borrower
EmploymentStatus, CreditScore, AmountDelinquent, BankcardUtilization. Also, we will look at his status with Prosper.com, what his ProsperScore is,
Term and ListingCreationYear.
Did you create any new variables from existing variables in the dataset?
I created two variables:
- ListingCreationYear: This is the year the loan was created, to consolidate
loans as they are currently listed in full date. - BorrowersGroups: This is the variable meantioned earlier that groups
borrowes into two groups, where Current, Completed, Past Due (1-15 days) and Past Due (16-30 days) are Non-Delinquent and the rest are delinquent borrowers.
Of the features you investigated, were there any unusual distributions?
Did you perform any operations on the data to tidy, adjust, or change the form
of the data? If so, why did you do this?
Knowing that every observation is a loan record, I run the below code to
notice that there are some duplicated that needs to be removed.
length(prosper_loans$ListingKey)## [1] 113937length(unique(prosper_loans$ListingKey))## [1] 113066# Removed all the duplicate loans listings
prosper_loans <- prosper_loans[!duplicated(prosper_loans$ListingKey), ]
length(prosper_loans$ListingKey)## [1] 113066## [1] 113066 17Ended up with 17 variables and 113,066 observations to be investigated
Bivariate Plots Section
Looking at the correlation coofecient in the relationship between every
varaible we are investigating in Prosper, we can see that some variables like Borrower APR and Borrower Rate are highly correlated but if we think of it, we know for a fact that these are identical variables and the correlation is due
to the fact that they are identical. We can also see a high correlation
between loan amount and the monthly payment which makes sence as the payments
increase with the size of the loan. We can also see a rather strong negative
correlation between the Borrower AP and the Prosper Score which can be a start
to our hypothesis regarding that Prosper gives high APR scores to risky
customers.
In trying to understand the variation of the APR between borrowers, I
started with the delinquent vs non-delinquent and its obvious that the
delinquent group pages on average a much higher rate of 25% vs 20% for the non-delinquent group and since APR rates are imposed at the beginning of the
loan, this gets us to guess the prosper might have believed that these
borrowers are riskier.
This make sense because we know that borrower rate and APR are complimentary
variables, but what we dont see is a perfect correlation between the 2
variables, for example a borrower can pay a rate of 20% and an APR between
20% and 30% which is quite a difference. We can also spot some outlier paid
borrower rate less than 10% but an APR of almost 22%.
Looking on how the APR changed over time, we see that the mean was high
then dropped then reached its peak in 2011 then dropped to its lowest rate ever in 2014. I have a hypothesis for it and it has to do that during 2008 and 2011 where the increase in APR is visible, due to the housing crisis the
borrowers where riskier, for that Prosper had to increase its APR to protect
its investors and maybe even attract investors as these where tough times.
I expected this plot to have a linear relation, but what I can see is that
a borrower can of an amount can have a huge variation in his/her APR.
We can vertical lines due to the nature of the loan amounts as they are usually
complete numbers like 20,000 or 10,000 etc but the numbers between them I
guess is due to the fact that the loan was not fully funded by investors.
For example, if you want to borrow $25,000 you might pay as low as 5% and as
high as 35% for the same loan amount
From this plot we can see that the 3 terms have very similar median, and
even in the same term we can see that the APR can range from 15% to 30% on the
36 month term (the most popular term of 80,000 borrowers)
The Prosper rating is the rating given day one for the loan, which I am not
sure if it affects the APR or vise versa but what I can see it a strong
relationship between this categorical variable and the APR, because as the loan becomes riskier(AA least HR highest) the average APR rate jumps from under
10% to as high as 35% although this graph is full of outliers. If we look at the lowest APR rate for HR (riskiest borrowers) we can see an APR
of 20% which is very similar to the median of the much less risky “B” rated
borrowers.
One of the most obvious variables will be a borrower salary, and it is as
anticipated, borrowers of low salaries have a median of 25%, while we can see
that the higher a person salary the lower he get in APR on average, but I am
again suprised with this outlier of 100,000+ salary who had an APR of 40%.
We can see this borrower here, he has a salary above 1,500,000 per month and unless Bill Gates is asking for a prosper loan, I believe this is a user entry error or maybe a data wrangling issue. For that I will mutate salaries above 250,000 and replot the above.
Changing salaries above 200,000 by dividing them by 100
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0 3199 4667 5565 6816 185082Plotting the monthly salaries after doing the above treatment that removed most of the outliers
Since the data is concentrated to the left, I experimented it with log10 and
now the data plot is more clear, we can’t still see the relationship between
the monthly salary and the APR levels, but its now clear that the majority of
borrowers are in the same financial situation with average salary of 5,000.
We can see a 2 vertical lined on zero and near zero, and I guess this is either
missing data or non-employed borrowers. We can see that themost popular APR
is of 35% as its spread along the monthly salaries from 100 to 12,000.
Expected a linear relation here as well, but I can conclude that most borrowers
are under 1% DTI and some are in the 10% DTI but lets replot it with log10 to
clear the congestion of points on the left
Similar to amount borrowed we can see that DTI is plotting vertical lines
that represent the ratio, which is highly concentrated between 0.2 and 0.3
but here we can’t say that DTI is pushing the APR high, because on the same
DTI, for example 0.01, we can see APR ranging from 5% to 40% and the same on
almost every DTI level.
DTI interquartile range
## Min. 1st Qu. Median Mean 3rd Qu. Max. NA's
## 0.000 0.140 0.220 0.276 0.320 10.010 8472
From this plot we see little evidence that the utilization of the bank card
have any effect on the borrower APR because as we can see that the APR ranges from 5% to 40% on every single bank card utilization ratio, but we can also
notice that there is almost little borrowes above the
summary(prosper_loans$MonthlyLoanPayment)## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 0.0 130.9 217.4 271.9 370.6 2251.5With 90% of loan payment between $130 and $370 per month, we can see that there
is something going on in this plot, we dont see straight lines as we did in
other variables and we can see on higher monhtly payments the APR doesnt go
beyond 25% although we cant confirm because data there is little.
In the amount delinquent for borrowers we don’t see a relationship with the borrower APR.
We can see some states median dropping below 20% and some above 20% but in
general we dont see extreme APR rates in any of the states and those that might
look low are mainly due to the smal number of loans so its difficult to reach
conclusions.
Looking at the 3 terms of loans provided, we can see a relationship between
the size of the loan and the term of the loan which makes sense because
usually bigger loans need more time to be paid which explains the difference
in median which is $5,000 for the 36 month term and more than double $11,000.
When comparing the delinquent borrowers to those whom are non-delinquent, we can see that delinquent borrowers median loan amount is $5,000 compared to
around $7,000 for the the non-delinquent. We can also notice that the
delinquent borrowers amount borrowered are less spread than non-delinquent with shorter tail and even the outliers are much conservative than the
non-delinquent group.
After removing both the NA salaries and the zero salaries,
we can see that the high earners have the lowers median APR of 18% and even
the range of APR max is lower than that of the not-employed and lower salaries.
We can see that the median of APR is dropping as salary increases assuring that a person’s financial security is key to him or her getting a lower
or higher APR.
cor.test(prosper_loans$BorrowerRate, prosper_loans$BorrowerAPR, method = c("pearson"))##
## Pearson's product-moment correlation
##
## data: prosper_loans$BorrowerRate and prosper_loans$BorrowerAPR
## t = 2337.9, df = 113040, p-value < 2.2e-16
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## 0.9896979 0.9899341
## sample estimates:
## cor
## 0.9898167We can see the strong correlation between the 2 identical variables Borrower
Rate and APR
cor.test(prosper_loans$ProsperScore, prosper_loans$BorrowerAPR, method = c("pearson"))##
## Pearson's product-moment correlation
##
## data: prosper_loans$ProsperScore and prosper_loans$BorrowerAPR
## t = -261.44, df = 83980, p-value < 2.2e-16
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## -0.6735593 -0.6661019
## sample estimates:
## cor
## -0.6698475We the negative correlation between a Borrower Prosper rating and the APR
he or she gets because as the rating goes higher the APR decreases for low
risk customers
Bivariate Analysis
Talk about some of the relationships you observed in this part of the
investigation. How did the feature(s) of interest vary with other features in
the dataset?
We can’t see a clear relationship between most variables and the APR rate,
but we know that Prosper tends to increase the borrowes APR if they think they are risky borrowers because when grouping delinquent and non-delinquent
borrowers we noticed that there is 5% more APR on average between the 2 groups.
This is clear when looking at the Prosper Rating given to the borrower day 1
and his/her APR.
Also, we can see a relationship between the borrower APR and Income range,
although couldn’t see it with monthly salary but in the income range its clear that the higher the salary the lower the APR.
Most important when it comes to the risky borrowers get a higher APR is in the
APR distribution over years, because when the economy was bad, we saw high
APR levels that went back to normal during the years after 2009. Also, we can see that Interest rate has 99% correlation but only because these
are almost the same variables so we can’t consider it a correlation to be
studied.
Did you observe any interesting relationships between the other features
(not the main feature(s) of interest)?
We can see a relationship between the size of the loan and the loan term,
because usually bigger loans tends to need more time to be paid but more
interesting is that Prosper not only charged the delinquent borrowers with
higher APR, they even gave them loans of smaller amounts, because as I noted in the above plot we can see that delinquent borrowers median loan amount is
5,000 compared to around 7,000 for the the non-delinquent.
Also we can see a rather strong negative correlation a borrower APR and the
Prosper rating which proves that APR is used by Prosper to protect themselves
against risky customers.
What was the strongest relationship you found?
The strongest was between the Borrower APR and the Intereste Rate, but also
we can see a relationship between the borrower status and the APR and also
between his/her status and the size of the loan they will borrow and
definitely the APR levels go down as salary increases making the borrower seem
more secure.
Multivariate Plots Section
##
## Not employed $0 $1-24,999 $25,000-49,999 $50,000-74,999
## 806 621 7241 31940 30749
## $75,000-99,999 $100,000+
## 16780 17188There are lots of things we can see in this plot, we can see that from 2013,
prosper stopped accepting not-employed borrowers whom APR was the highest among all groups. We can see that after 2011, APR levels for all groups started getting lower to reach its lowest levels in 2014.
This is amazing, we can see very clear that Prosper penalizes risky borrowers
with high levels of APR, all over the plot HR (High Risk) is never visible
until we reach 35%, and I believe these are the not-employed introduced in
2013.
lets zoom in on the levels to see how they are distributed
After zooming in, we can see that almost some levels don’t overlap, the “AA”
sits on the left side with most of the borrowers getting less than 10%,
then the “A” between 10% and 15% then the B between 15% and 20%.
The difference is clear between the borrower APR and Interest rate between
delinquent borrowers and non-delinquent, so if Prosper think you are riskier
you will get a higher APR.
When plotting using confidence ellipses we can see the seperation between
the different groups of borrowers rating, its clear that the low rik “AA”
borrowers almost dominate the lowest Borrower Rate/APR an then the graph
goes through different groups accordingly, we can see that the more a borrower
is flagged as risky, the more we can see the groups intersect. We can also
notice that the highest risk borrower “HR” is a very small group almost not
visible on the plot.
We can see lots of interesting things here, Its now clear that Prosper borrowers whom ended up to be delinquent where in the first place risky as
they have the majority of the skyrocket APR levels. We can also see that the 60 month term borrowers had a lower APR than the popular 36 month term
borrowers in both groups.
We know that the bulk of the borrowers are from years 2011 till 2013, for that
the above years are only visible in the plot.
Looking at the years after the financial crisis we can see the difference
in the distribution of the Borrower APR expecially in 2013 when it started
to become normal where almost half of the loans are above 20% and half
almost below unlike 2011 where we see spikes in the density at 30% and 40% APR and same for 2012 but less moderate.
I was reluctant to add 2014 because the data is not complete but if we look at what data we have now, it shows that APR dropped and is becoming skewed to the
right where the majority of loans are in the 15% APR range which is either
due to Prosper attracting people with better ratings or the general economical
situation is recovering.
We can see how much APR levels has dropped if we look, we can
see that the are no more outliers in 2014 and we can see that even the HR
rated borrowers dont exceed at max the 30% APR, we can also notice that the IQR was big on every level but now in 2014 it has shrinked
——————————————————————————-
Multivariate Analysis
Talk about some of the relationships you observed in this part of the
investigation. Were there features that strengthened each other in terms of
looking at your feature(s) of interest?
It is clear more that a person income range affects the APR he or she get at
Prosper. We can also see that the borrower rating affects the APR which
gets us to conclude that the difference between the APR and interest is mainly
due to the borrower riskiness.
Were there any interesting or surprising interactions between features?
The interesting part was the APR levels dropped over the years which is
something quite odd especially with financial institutes whom tend to usually
add fees and interest rates as they grow bigger.
The surprising was the introduction of the not-employed people in 2013 with
skyrocket APR.
Final Plots and Summary
Plot One
Description One
I chose this to be the first because it might have explained maybe most of the
variation in APR that is not explained by the interest rate. This here I
believe affects the initial prosper rating. It is clear when the APR levels starts to go very high when the borrower monthly salary drops which can be
double between a high earner and someone denoted as not-employed which allows us to confirm that the financial security of a borrowers highly influences
his or her APR level.
Plot Two
Description Two
This plot was very supporting because as we can see that the prosper rating
given day 1 to the borrower affects the APR level he or she get, we can almost
see exclusivity for the borrowers rating levels as if they are separate bin.
For example the AA ratings never go beyond 10% in APR while the C rating sits
between the 15% and the 20% as for the HR, we now know that these are
“not-employed” borrowers from 2013 whom where mostly given APR levels of 35%
as they are risky and might become delinquent.
Plot Three
Description Three
I wanted to concentrate on this because it was surprising that the APR level
dropped and is still dropping at Prosper, I believe this can be due to many
reasons but my theories are that probably Prosper is attracting less riskier
highly paid borrowers (although they gave loans to non-employed borrowers in
2013) and due to the recovery of the US economy after the housing crisis in
2008-2009.
Reflection
I was interested in this dataset because I was interested in Prosper as a company but when looking at the number of variables and since there are 81, I was reluctant at first because there are too many to be analyzed and even
choosing few variables wasn’t easy, I had to repeat the project 3 times before finding a problem that was unique, interesting and which I can choose adequate
number of variables but not too many.
Also the plotting wasn’t easy as the size of the data is big, and mostly was skewed which caused plots to be useless, even I was not satisfied with with
ggpair plot because the variables that I chose was too much to be plotted at
once.
Maybe I should have studied the riskiness of the borrower rather than the
APR difference. It started with me because I thought APR was the same for all
borrowers, but when I came across the difference in APR I wanted to know why,
to be honest I was dreaming of this highly correlated continuous variables that
will popup from somewhere and just explain it all but it seems in real life EDA isn’t straight forward.
I believe I have concluded that the borrower salary has
a huge affect on the APR he or she gets but I don’t think this alone helps
because as we can see APR levels dropped over the years. I believe Prosper has system that measures the borrower riskiness because as
we saw over and over again, borrowers who are delinquent tended to have higher APR levels.
Now when I look at the variables and the prosper dataset I feel very familiar
with it and I now have an idea how loans work, I know what affects the amount
or the APR levels, also I know that its not only for low income borrowers, I
also know that it has dramatically changed over time and that the average
borrowerd amount is around $7,000.
Maybe if we add how salaries changed over the years, this could have been quite
an idea, because there we can know if Prosper is attracting more secure
borrowers or is it the economy, also if we can add a benchmark for what APR
levels other P2P companies are providing and even what bank are providing in
terms of micro loans.
I think there are lots of issues that can be investigated and even after
I feel I left lots of issues that could have been investigated especially how Prosper rates its borrowers upon approval of the loan which came late to me
when I was almost done with the APR analysis and couldn’t go back to start
again for now.